1. Field of the Invention
This invention relates generally to multicolor in situ hybridization methods that detect genetic abnormalities. In particular, this invention relates to novel methods to prepare cells for multicolor in situ hybridization and to detect genetic deletions, multiplications, and chromosomal haplotypes associated with genetically related diseases.
2. Brief Description of Background Art
Over 2000 human diseases result from DNA alterations including deletions, multiplications and nucleotide substitutions. Finding genetic disease alterations in individuals and following these alterations in families provides a means to confirm clinical diagnoses and to diagnose disease in carriers, preclinical and subclinical affected individuals, affected unborn fetuses, and preimplantation embryos. Counselling based upon accurate diagnoses allows patients to make informed decisions about potential parenting, ongoing pregnancy, and early intervention in affected individuals.
Disease associated deletions, multiplications, and substitutions may be large or small. Diseases such as steroid sulfatase deficiency, alpha-thalassemia, and Duchenne muscular dystrophy usually result from large deletions. Steroid sulfatase deficiency and alpha-thalassemia patients have common deletion breakpoints while Duchenne muscular dystrophy patients have many deletion and duplication breakpoints. Diseases such as Charcot-Marie-Tooth Disease Type 1A (CMT 1A) have multiple duplication breakpoints that cause the same disease by multiplying the same specific gene(s).
Even in the absence of deletions or multiplications, nucleotide substitutions can result in genetic disease. Multiple alleles within genes are common. Most alleles result from neutral mutations that produce indistinguishable, normally active gene products or express normally variable characteristics like eye color. In contrast, some alleles are associated with clinical disease like sickle cell anemia. Many disease associated alleles co-segregate with specific haplotypes reflected by a unique sequence of neutral nucleotide changes in that gene region. Identifying and following the segregation of affected haplotypes in families provides a means to identify a disease-related gene even when the specific gene mutation is unknown.
Previously diagnosis and confirmation of genetic disease and carrier states often relied upon enzyme activity testing, statistical analysis, or invasive diagnostic procedures. For example, painful nerve conduction tests have been necessary to detect preclinical and subclinical cases of CMT1A, and amniocyte steroid sulfatase assays have been used to prenatally diagnose steroid sulfatase deficiency.
DNA polymorphisms (RFLPs; Restriction fragment length polymorphisms) have been used to diagnose more than 20 genetic diseases reliably (See Lebo et al, Am. J. Hum. Genet. 47:583-590, 1990a, for list). The DNA Committee of the Pacific Southwest Regional Genetics Network proposed that a prenatal clinical test must be informative in at least 70% of fetuses and must be at least 95% reliable (Ibid.). All subsequent communications have agreed with these criteria. The percent informative matings are calculated according to Chakravarti and Buetow, Am. J. Hum. Genet. 37:984-997 (1985), with different formulas for autosomal recessive, autosomal dominant, and X-linked genetic disease. Not all matings are informative because parents may be homozygous for neutral DNA polymorphisms (FIG. 1). The proportion of informative matings depend upon 1) the number of different alleles at each gene locus, 2) the relative frequency of each allele in the population (the most informative have more than one common allele), and 3) whether alleles are distributed randomly throughout the population. Finding enough informative polymorphisms can be very laborious when few or uncommon polymorphisms are found at a disease locus [Lebo et al, Am. J. Hum. Genet. 47:583-590 (1990]. Using characterized polymorphisms may be laborious since several often need to be tested. See, e.g., Lebo et al, Am. J. Med. Genet. 37:187-190, (1990). Even then a proportion of uninformative results in some pedigrees are anticipated. For example, although a substantial number of polymorphic loci have been reported in the CMT1A gene region, multiple polymorphic tests must be used to identify carriers and affecteds. Further complicating accurate diagnosis is the low but real recombination frequency between the reported flanking polymorphisms and the CMT1A locus. Thus for diseases like CMT1A a direct DNA test would provide a more rapid, reliable, and tolerable diagnostic test.
In situ nucleic acid hybridization (in situ hybridization) is the most direct means to map unique chromosome-specific sequences. Originally in situ hybridization used radiolabeled probes to detect target genes [Harper and Saunders, Chromosoma, 83:431-439 (1981)]. While radiolabeled probes detected unique nucleic acid sequences on metaphase chromosomes, detection lacked both speed and resolution. Fluorescent labels conjugated to DNA probes localized targets quickly with high resolution [Langer-Safer et al., Proc. Natl. Acad. Sci. USA, 79:4381-4385 (1982)] to identify single copy genes in metaphase chromosomes [Landegant et al., Nature, 317:175-177 (1985)].
Simultaneously using different colored nucleic acid probes developed for repetitive sequences [Nederlof et al., Cytometry, 10:20-27 (1989)] ordered cosmid clones on a region of uniformly stained chromosome [Lichter et al., Science, 247:64-69 (1990)] and resolved closely spaced DNA clones in interphase nuclei [Trask et al., Genomics, 5:710-717 (1989)]. Different colored probes have been used to demonstrate chromosome translocation in chronic myelogenous leukemia in interphase nuclei [Tikachuk et al., Science, 250:559-562 (1990)].
While substantial nonspecific signal previously limited diagnostic utility, recent studies demonstrated that a 40kb myophosphorylase cosmid probe gave two unique hybridized signals on every interphase nucleus observed [Lebo et al., Hum. Genet. 86:17-24 (1990)]. This result demonstrated that in situ hybridization to interphase nuclei was sufficiently specific to count gene target number. This specificity has been demonstrated using a Texas red labeled 2kb keratin cDNA hybridized simultaneously with a yellow-green fluorescein labeled 3kb flanking genomic sequence that each identify two and only two targets in blue DAPI stained normal interphase nuclei (hereinafter referred to as multicolor in situ hybridization).
Recently protocols have been improved to order 2kb DNA targets with either Texas red label or yellow-green fluorescein label on blue DAPI banded metaphase chromosomes [Lebo et al., Hum. Genet. 88:13-20 (1991); Lebo et al., Am. J. Hum. Genet. 50:42-55, 1992]. This improved sensitivity and specificity makes it possible to map small gene probes to morphologically distinguishable chromosome regions. Closely spaced clones within 250kb have recently been ordered on banded chromosomes. Thus both copies of duplicated disease genes separated by about 250kb are anticipated to be resolved on early metaphase chromosomes to test for duplicated disease genes directly.
DNA replication that generates a double hybridization signal is difficult to distinguish from true intrachromosomal duplication [Lupsky et al. Cell, 68:219-232 (1991)]. Efforts to minimize cells with replicated gene targets have included using confluent amniocyte and fibroblast cultures that have reached stationary phase as well as testing direct amniocytes and chorionic villus cells.
Protocols allowing different cells to be tested by multicolor in situ hybridization would provide extended clinical applicability of this protocol. False negative in situ hybridization results must be minimized by optimally processing and fixing cells to slides according to cell type and slide age. Optimal proteinase K digestion removes histone and non-histone cellular and chromatin-bound proteins to expose nearly all gene sequences to probe without removing the remaining protein cytoskeleton that holds the nuclear DNA in place. Excessive and under digestion each result in failure to detect gene sequences (false negatives). The same multicolor in situ hybridization protocol that optimally detects lymphocyte genes will detect no genes in cultured fibroblasts. Likewise protocols that work on fresh preparations will detect no sequences on old slides because proteolytic enzyme digestion must be increased to expose unique DNA sequences.
Although in situ hybridization has not been applied to prenatal diagnosis previously, scoring each hybridized cell individually can distinguish common fetal cells from contaminating maternal cells in prenatal samples to give an unambiguous gene deletion or multiplication diagnosis. In contrast, contaminating DNA from nucleated maternal blood cells mixed with direct amniocytes or maternal decidual cells in improperly dissected chorionic villi can be amplified along with fetal DNA by the polymerase chain reaction (PCR) to give false gene signals. Currently restriction enzyme analysis is used to obtain reliable results on DNA extracted from direct amniotic fluid samples from at-risk alpha-thalassemia fetuses so that minor contaminating maternal bands which are significantly lighter do not interfere with diagnosis [Lebo, Hum. Genet. 85:293-299, (1990)]. Culturing all tested amniocytes removes maternal blood cells but increases diagnostic time 2-3 weeks. Excellent dissection of chorionic villi will remove nearly all maternal decidual cells, but is technically difficult and requires specialized training. PCR results cannot be trusted with suboptimally dissected chorionic villi because maternal decidual cells grow along with fetal chorionic villus cells. Thus multiple samples may have to be taken when fetal and maternal DNA results do not differ. In contrast, testing cells directly by in situ hybridization can distinguish a small contaminating maternal cell population among the fetal cells because each cell is scored individually. In addition smaller maternal lymphocyte nuclei can be distinguished visually from the larger direct amniocyte nuclei. Thus multicolor in situ hybridization eliminates culture time and false results from maternal cellular DNA amplified indiscriminately by the polymerase chain reaction.
Prior to the present invention, identifying the specific haplotype of a chromosome was cumbersome and time consuming. Chromosomal haplotypes contain many mutations and probing a sufficient number for familial segregation analysis is labor intensive. Despite the many mutations present in a haplotype, the variability between individuals is expected to be small because the races are considered to have diverged about 200,000 years ago based upon no differences in the alpha-globin gene exons in the three races. The intron and gene flanking sequences are expected to have greater variability. Recently HLA intron sequences have been analyzed and homology on the order of 95% has been found. The FCG2 genes which share 98% exon homology have been distinguished from the FCG3 genes which share 98% homologous sequences but differ from the FCG2 genes by about 5% [Lebo et al., Hum. Genet. 88:13-20, (1991)]. Prior hybridization protocols were unable to distinguish this level of homology. Haplotype analysis by in situ hybridization is easy and direct even in prenatal samples where maternal contamination can confuse the results.
The present invention overcomes these and other shortcomings by providing a novel combination of multicolor in situ hybridization and analysis of geometric fluorescence patterns to increase the proportion of nuclei correctly scored for genetic aberrations. Methods of multicolor in situ hybridization are provided to allow precise and rapid genetic haplotype analysis. Novel methods of cellular preparation expand the applicability of the claimed methods to prenatal diagnosis.